没有 crosstab/tablefunc 的枢轴 table

Pivot table without crosstab/tablefunc

我有一个table这样的

输入

id    author    size    file_ext
--------------------------------
1     a         13661   python
1     a         13513   cpp
1     a         1211    non-code
2     b         1019    python
2     b         6881    cpp
2     b         1525    python
2     b         1208    non-code
3     c         1039    python 
3     c         1299    cpp

我希望能够按以下方式旋转此 table

输出

id    author    size    python    cpp    non-code
-------------------------------------------------
1     a         13661   1         0      0
1     a         13513   0         1      0
1     a         1211    0         0      1 
2     b         1019    1         0      0
2     b         6881    0         1      0
2     b         1525    1         0      0
2     b         1208    0         0      1
3     c         1039    1         0      0
3     c         1299    0         1      0

我能在网上找到的所有文章 pivot tables 基于第二列。我的最终目标是为每个 ID 获取一条记录。

最终输出

id    author    size    python    cpp    non-code
-------------------------------------------------
1     a         28385   1         1      1
2     b         10633   2         1      1
3     c         2338    1         1      0

此处汇总了大小、python、cpp、非代码列的值。

有条件聚合:

select 
  id, author,
  sum(size) size,
  sum((file_ext = 'python')::int) python,
  sum((file_ext = 'cpp')::int) cpp,
  sum((file_ext = 'non-code')::int) "non-code"
from tablename
group by id, author

参见demo
结果:

> id | author |  size | python | cpp | non-code
> -: | :----- | ----: | -----: | --: | -------:
>  1 | a      | 28385 |      1 |   1 |        1
>  2 | b      | 10633 |      2 |   1 |        1
>  3 | c      |  2338 |      1 |   1 |        0

您还可以使用名为 tablefunc 的 PostgreSQL 交叉表函数扩展。我建议您看一下这个 link(给出的示例与您想要的结果非常相似):https://vertabelo.com/blog/creating-pivot-tables-in-postgresql-using-the-crosstab-function/

虽然您想避免使用 crosstab() 函数,但使用聚合 FILTER 子句以获得最佳性能和最清晰的代码:

SELECT id, author
     , sum(size) AS size
     , count(*) FILTER (WHERE file_ext = 'python')   AS python
     , count(*) FILTER (WHERE file_ext = 'cpp')      AS cpp
     , count(*) FILTER (WHERE file_ext = 'non-code') AS "non-code"
FROM   tablename
GROUP  BY id, author;

这是仅使用聚合函数的最快方法。参见:

为了绝对最佳性能,crosstab() 通常更快 - 尽管在这种情况下更冗长:

SELECT id, author, size
     , COALESCE(python    , 0) AS python
     , COALESCE(cpp       , 0) AS cpp
     , COALESCE("non-code", 0) AS "non-code"
FROM   crosstab(
$$
SELECT id, author
     , sum(sum(size)) OVER (PARTITION BY id) AS size
     , file_ext
     , count(*) AS ct
FROM   tablename
GROUP  BY id, author, file_ext
ORDER  BY id, author, file_ext
$$
, $$VALUES ('python'), ('cpp'), ('non-code')$$
) AS (id int, author text, size numeric
    , python int, cpp int, "non-code" int);

同样的结果。

db<>fiddle here - 有中间步骤.

详细解释:

  • PostgreSQL Crosstab Query
  • Pivot on Multiple Columns using Tablefunc

关于聚合函数(sum(sum(size)) OVER (...))上的window函数,参见:

  • Get the distinct sum of a joined table column

如果同一个 id 应该有多个 author,请注意细微差别:在这种情况下,第一个查询 returns 多行,crosstab() 变体只选择第一作者。